A Yearlong Ethiopian Rural Household Survey YLERHS is utilized for this study. The survey is conducted by the Economics Department of Addis Ababa University in collaboration with the USAID/Ethiopia. The YLERHS is an extension to the 5th round panel data of the Ethiopia Rural Household Survey ERHS that covered 1685 households from 18 villages across four regions Tigray, Amhara, Oromiya and SNNP . It collected information for a typical period, based on households recall either “for the last week” or “for the last four months” prior to the survey period. The YLERHS, however, purposely selected 4 villages Yetmen and D/Brihan from Amhara region, Eteya from Oromiya region and Azedebo from SNNP region and extended field work for the period of more than one year; April 2000 to June 200161F57. These villages were selected to represent different agro‐ecologies as well as crop diversifications: teff Yetmen , wheat Eteya and perennials crop Azedebo and animal husbandry as integral to farming system D/Birhan . A sample of 247 households62F58 62 households from each village who had been interviewed in the 5th round were re‐interviewed every fortnight to record the transaction as it was rather than based on recalls.
The YLERHS has detailed information on socio‐demographic, consumption, crop output, livestock ownership, and non‐farm and off‐farm activities, among others. Table 1 presents major crops grown in the villages, percentage of households growing those crops and output in qt/ha 63F59. Households in Yetmen 93 percent and Azedebo 31 percent are growing teff. For positive producers, average teff production was
57 The survey launched in the month of April just to align with the beginning of the first plough.
58 One household is dropped due to incomplete information.
59 It is the actual measured quantity of output. During survey periods land cultivated and crops output was measured rather than depending only on the responses of household heads. The average is computed for positive producers.
157 9.21qt/ha in Yetmen major teff growing village and 4.5qt/ha in Azedebo. All households in D/Brihan and about 23 percent in Eteya are growing barely and produced about 14qt/ha. Wheat is cultivated in all villages that accounts for 30 percent of households in Yetmen and Azedeob, 84 percent in D/Brihan and 95 percent in Eteya.
On average, about 19qt/ha of wheat was produced in Eteya main wheat growing village , 9.5qt/ha in D/Brihan and 8qt/ha in Yetmen and Azedebo. Households growing maize account for 47 percent in Yetmen, 14 percent in Eteya, and 37 percent in Azedebo. The average yield is limited to 5.35qt/ha in Yetmen and about 9qt/ha in Eteya and Azedebo. Sorghum is growing by less than 10 percent of households in D/Brihan and about 20 percent in Eteya, with average yield responses of 10qt/ha and 13qt/ha, respectively.
Table 1: Percentage of households growing crops and average yield produced by households growing
D/Brihan Yetmen Eteya Azedebo
Crop % of hh growing crops
Quantity produced
in qt/ha
% of hh growing
crops
Quantity produced
in qt/ha
% of hh growing
crops
Quantity produced
in qt/ha
% of hh growing crops
Quantity produced
in qt/ha
Teff ‐ ‐ 93.4 9.2 3.4 ‐ ‐ 30.7 4.5 3.2
Barely 100 13.4 5.8 ‐ ‐ 22.6 13.5 9.5 ‐ ‐
Wheat 83.9 9.5 5.4 29.5 7.8 3.1 95.2 18.4 7.3 30.7 8.3 4.4
Maize ‐ ‐ 47.5 5.5 3..3 14.5 8.4 3.3 37.1 8.2 7.1
Sorghum 8.1 10.5 4.6 ‐ ‐ 21.0 12.6 5.7 ‐ ‐
Bean 74.2 11.4 8.2 ‐ ‐ 37.1 11.5 2.5 6.5 3.8 1.6
Vetch ‐ ‐ 41.0 8.1 5.6 ‐ ‐ ‐ ‐
Source: Authors’ computation from survey data Figures under parentheses are standard deviations.
Households annual gross income is derived from crops output value of total output , livestock/products sales, wage & self‐employment as well as transfer income. On average, households earned in per capita about Birr 880.8 in Eteya, Birr 672.4 in D/Birhan; Birr 477.2 in Yetmen and Birr 114.1 in Azedebo Table 2 . The largest share of these income is obtained from crops output in three of four villages; 69 percent in
158 D/Brihan, 85 percent in Yetmen and 82 percent in Eteya. In Azedebo, crop income accounts for only 29 percent; transfer income has the largest share 36 percent Table A1 . Livestock/product sales account for the second and third largest share of income in D/Brihan and Azedebo, respectively Table A1 . Wage & self‐employment accounts for 13 percent of income in Azedeob albeit less than 5 percent in other villages Table A1 .
About 80 percent of households reported crop output sales at least for one crop Table A1 . Teff in Yetmen and wheat in Eteya are the two major crop output sales reported by more than 95 percent of households. This could be implying that these crops are mainly grown for cash requirement of the households as they are fetching better prices in the markets. Output sales of “other crops” are reported by about 89 and 86 percent of households in Eteya and Azedebo Table A2 . In general, sales account for 55 percent of total crop output in Azedebo, 36 percent in Eteya and 21 percent in Yetmen but only 5 percent in D/Brihan Table 2 . As cash crop such as coffee and chat are growing in Azedebo, the sales account for the largest share of crop output in the region.
Table 2: Sample mean of income, consumption expenditure and related variables by villages
D/Brihan Yetmen Eteya Azedebo
Per capita annual income in Birr 672.4 477.2 880.8 114.1
270.3 259.6 566.8 111.2
Share of income from crop output in % 69.2 85.3 82.3 29.1
Share of crop output sales in the total output 4.7 21.2 36.8 54.8 Per capita annual food expenditure in Birr 505.1 385.3 405.8 168.3
144.3 162.8 143.1 61.7
Food expenditure share in % 81.2 83.5 62.3 77.7
Age of household head in years 52.0 47.1 45.9 45.6
Sex of head; 1 if male 0.8 0.8 0.9 0.8
Family size 6.0 5.3 7.7 7.3
Source: Own computation from survey data.
159 Households were requested to report total quantity of food consumption from own harvest, purchase and received as gift either from relatives or other sources . Consumption expenditure is also reported for purchased food items, which allows us to derive unit value as a ratio of consumption expenditure divided by quantities of consumption. The unit value price is used to impute consumption expenditure from own harvest as well as received from gift. Total quantity of consumption is also converted into calorie consumption based on Ethiopian Health and Nutrition Research Institute EHNRI calorie conversion factors. Total expenditure and calorie consumption are aggregated at household level and standardized into per capita expenditure and daily calorie consumption, respectively.
Household per capita food consumption was, on average, about Birr 405.8 in Eteya, Birr 505.1 in D/Birhan, Birr 385.5 in Yetmen and Birr 168.3 in Azedebo. Share of food in the total household budget is also computed to account for the importance of food in the total household budget. Food accounts for about 80 percent of household budget in three of four villages; in Eteya it accounts for 62 percent. Table 3 presents the share of food by seasons lean, harvest and post‐harvest . Food accounts for about 70 percent and 75 percent during harvest Dec.‐ Feb. and post‐harvest March‐ May seasons, respectively, but about to rise to 80 percent in lean seasons June‐August as well as September‐November Table 3 . In harvest as well as post‐harvest seasons, non‐food expenditure accounts about 30 percent of household budget. This is presumably due to the fact that households often postpone most of their non‐food demand to these seasons, as crop outputs are the major sources of rural income that could be used for various purposes. In the lean season, on the other hand, the larger share of the retained income is solely devoted for food consumption.
160 Table 3: Food budget shares by seasons in %
Jun.‐Aug. Sep.‐Nov. Dec.‐Feb. Mar.‐May
Food share 79.3 79.8 68.1 74.1
Cereals 51.9 54.2 49.9 48.7
Teff 24.2 19.7 20.9 27.2
Barley 14.6 16.0 22.5 18.9
Wheat 24.0 24.5 24.1 24.1
Maize 29.0 32.2 23.5 23.1
Sorghum 4.7 3.6 3.7 3.6
Other cereals 2.2 2.7 2.4 2.7
Pluses 15.5 13.7 13.3 14.2
Horse bean 54.0 72.4 69.9 59.9
Other pluses 46.0 27.6 30.1 40.1
Root crops 10.7 12.1 11.5 10.5
Enset 27.5 35.1 26.6 22.0
Potato 72.5 64.9 73.4 78.0
Fruits & vegetables 0.9 0.9 2.2 0.9
Animal products 7.0 4.8 6.1 9.9
other foods 13.1 12.6 15.8 15.8
Coffee 21.7 21.3 22.7 21.9
Sugar 10.9 10.3 8.6 9.0
Cooking oil 16.4 16.4 13.8 14.7
Others 50.2 51.2 51.6 53.2
Source: Authors’ computation from survey data
Cereal teff, maize, wheat, barley and sorghum are major staple crops account for more than 50 of food budget Table 3 . Maize has the lion share that ranges from 29 percent to 33 percent of cereal consumption in the lean seasons, followed by teff and wheat about 24 percent . During harvest season, wheat, barley and maize accounts for about 23.5 percent of cereal consumption. Teff consumption dominates household food budget in April‐February 27.2 percent . This could be presumably due to the fact that teff threshing has taken place in these months often after all crops as it need more intensive care and also because of the Easter Holiday that colorfully celebrating in the month of April. During Easter Holiday, most Ethiopian families are making prestigious pancake food, locally known as Injera, made from fermented teff flour.
In general, Injera is the Ethiopian staple bread which is also made from barely, wheat, maize, sorghum flour or a mixture of teff flour with others. While most of urban and
161 better‐off rural households often use teff, majority of rural households use other cereals and/or a mixture with teff. The staple bread is often supplemented by sauces prepared from pluses, potato and animal products as well as fruits & vegetables. Pulses accounts for about 13.7‐16.9 percent of food budget while fruits & vegetables for about 1.1‐2.3 percent and animal products ranging from 4.4 percent in June‐August to 11.9 percent in March‐May. Other foods such as coffee, sugar and cooking oil accounts for 16.5‐20.4 percent; coffee shares nearly a quarter of other foods share, as almost all households are commonly making and drinking coffee at least three times a day.
Animals products consumption are essentially increased during months coincides with the Holidays: the Ethiopian New Year on 11th of September , the commemoration of the Founding of True Cross on 26th of September , Christmas on 7th of January and Easter Sunday in April , among others. During these festivals animals are slaughtered to lavishly feed not only own family members but also visitors and relatives outside of the family. The New Year festival is more colorfully celebrated in the most part of the country. However, for Christian ethnic groups in the southern region, Meskel feast is the most important event of the year that lasts for entire week. Migrant workers return home, gifts are exchanged, new clothes are bought and a lot of feasting takes place.
Moreover, the consumption of animal products considerably increase in March‐May about 10 percent as the Easter Holiday is celebrating after 55 days of fasting and abstaining from eating meat and dairy products particularly the Orthodox Christian worship to commemorate the 40 days of fasting of Jesus experienced before crucifixion64F60.
60There are also other colourfully celebrated Muslim holidays in Ethiopia which leads to have seasonal impact on food consumption of households.
162
Table A3 presents food budget share by villages. Obviously, food accounts for the largest share of household budget and cereals consumption mount in each village.
However, the relative importance of these cereals varies among villages. Barley is more important food items in D/Brihan, teff in Yetmen and wheat in Eteya. Root crops, especially enset, are the second most important food item in Azedebo. The importance of crops in each villages largely dominated by the types of crops growing in the villages; for instance, barley is mainly growing in D/Brihan, teff in Yetmen, wheat in Eteya and maize in Azedeob Table 1 .
Food purchase accounts for about 30 percent of household expenditure, on average Table 4 . There are disparities among expenditure quintiles and villages, however.
Households in the lowest expenditure quintile rely on market for about 34 percent of their consumption; in Azedebo, it accounts for about 40 percent. The share eventually declines to 23 percent in the third quintile before it increased to 27 percent in the fourth and further to 33 percent in the fifth quintiles. It is implying that the poor as well as better‐off households more likely rely on market apparently for consumption shortfall or household consumption preferences than households in the middle income groups. Moreover, the share of purchase in total consumption eventually increases across income quintiles in three of four villages.
163 Table 4: Food purchased share in total expenditure by villages and expenditure quintiles in %
Q1 Q2 Q3 Q4 Q5 Total
All 33.9 26.5 22.7 26.6 32.9 28.5
Debre Birhan 15.1 21.4 20.4 24.6 20.7 22.4
Yetmen 16.0 22.1 24.5 26.2 ‐ 22.3
Eteya ‐ 16.6 21.9 31.2 33.9 32.7
Azedebo 40.3 31.2 ‐ ‐ ‐ 36.6
Source: Authors’ computation from the survey data.
Note: ‐ means no household belongs to that specific income groups
Table 5 presents cereal purchase share in household food budget. Maize purchase accounts for the largest proportion 53 percent followed by teff 17 percent and wheat 12 percent . Among villages, while teff purchase dominates in D/Birhan 53 percent and Eteya 33 percent , maize does in Yetmen 79 percent and Azedebo 82 percent . Households in D/Brihan and Eteya depend on market for teff consumption, as they are not growing teff Table 1 . Maize is cultivated in Yetmen and Azedebo but less than half of households are reported growing. Accordingly, non‐growers and growers with consumption shortfall depend on the markets. The demand and supply of food items may also depend on household wealth status and relative prices of food items. The poor households produce teff and wheat mainly for marketing as they fetch better prices; and instead demanding maize and sorghum at relatively lower prices.
Households in Yetmen, for instance, produce teff and supply to distant market places such as Addis Ababa and instead purchase maize for consumption. For poor households, cheaper items could grant them to maintain more quantity of grain for consumption. Better‐off households, on the other hand, may be demanding teff and wheat. Similarly, households in Eteya and D/Brihan are more likely supply wheat and barley, respectively, and purchase teff. Households in Azedebo are producers and consumers of roots crops, especially enset. They are also producing maize for
164 subsistence consumption. In case of consumption shortfall, they do also purchase maize as it is cheaper than other cereals.
Table 5: Cereals purchased by villages in %
D/Birhan Yetmen Eteya Azedebo All
Teff 52.9 0.5 32.4 7.1 16.9
Barley 1.5 1.0 9.2 0.5 4.0
Wheat 2.9 7.0 14.7 9.6 11.4
Maize 0.0 78.5 10.8 81.5 52.6
Sorghum 10.7 0.9 29.9 1.3 12.7
Other cereals 32.0 12.0 3.0 0.0 2.4
Source: Authors’ computation from survey data
Further analysis by expenditure quintiles and food items revealed that larger portion of teff purchase accounts for households in the higher income quintiles while maize does in the lower income quintiles Table 6 . Of the total amount of teff purchase, for instance, while 25 percent and 52 percent is accounted for households in the fourth and fifth quintiles, respectively, it is limited to 10 percent for households in the two lowest quintiles and about 7 percent in the middle quintile. For households in the lowest and highest expenditure quintiles, wheat purchase accounts for the largest share. Wheat is commonly used for making bread, which equally serve both the poor and better‐off households. The share of wheat purchase is low for the middle income presumably they may use from their own harvest, rather than relying on markets.
There is no clear pattern for barley; it accounts for about 20 percent in the middle income and 69 percent in the highest income but less than 5 percent in the remaining quintiles. Enset purchase is largely dominated by households in the lower income quintiles. Households in the highest income quintiles more likely purchase animal products, fruits & vegetables, coffee, sugar and cooking oil Table 6 .
165 Table 6: food purchase by expenditure quintiles in %
Q1 Q2 Q3 Q4 Q5
Teff 10.5 10.8 0.7 25.8 52.1
Barley 5.6 1.0 19.4 4.9 69.1
Wheat 30.7 13.0 4.6 22.7 29.0
Maize 48.6 36.2 5.7 2.4 7.1
Sorghum 1.2 4.3 1.1 22.2 71.1
Other cereal 13.9 10.0 20.4 15.8 39.9
Horse bean 41.0 31.9 9.7 9.3 8.1
Other pluses 20.3 21.6 8.8 19.7 29.6
Coffee 2.9 13.0 19.3 29.5 35.4
Enset 77.0 19.7 0.4 0.3 2.6
Potato 19.6 19.8 11.2 17.9 31.5
Fruits & vegetables 8.3 3.7 5.2 35.2 47.7
Animal products 3.0 9.2 30.2 26.0 31.6
Sugar 0.5 4.7 12.9 36.4 45.5
Cooking oil 2.6 7.3 16.9 32.6 40.6
Other foods 10.6 13.9 19.4 27.1 28.9
Source: Authors’ computation from YLERHS data
Fig.1 presents daily per capita calorie consumption over the course of the year by villages. Among village, there is a considerable fluctuation in the level of calorie consumption; it is quite remarkable among households in D/Brihan Fig.1 . Apparently, the lowest level of calorie consumption is observed in Azedebo. In general, calorie consumption rises in the months that coincide with Holidays September and April as well as during harvest season December‐February . In the months of September the Ethiopia New Year Enkutatash and the Founding of True Cross Meskel festivals are colorfully celebrating on September 11th and 26th, respectively.
Figure 2 presents calorie consumption by expenditure quintiles. The level of calorie consumption increases with household expenditure quintiles as expected; households in the higher expenditure quintile consume more calories than households in the next lower income quintile. Moreover, the level of consumption proportionally increases, within each income quintile in the months of September, January and April. On average, about 1200, 2000 and 2500kcal/day/person is consumed for households in the lower, middle and higher income quintile, respectively.
166
Fig.3: Calorie consumption in kcal/person/day by months and villages
Fig. 4: Calorie consumption in kcal/person/day by months and income quintiles
4 Empirical Results
Table 7 and 8 present OLS estimates of log calorie consumption in kcal/day/person on monthly dummies controlled for household income and its squared, total livestock owned, household composition, age, sex and education level of head. While Table 7 presents the results for the 12 months January 2001 is a reference , Table 8 presents for seasons December‐February is constant . The Wald tests for joint significance of monthly/seasonality dummies are significant for all estimates except for Eteya under
0 500 1000 1500 2000 2500 3000
3500 D/Brihan Yetmen Azedebo Eteya
0 500 1000 1500 2000 2500 3000
3500 Q1 Q2 Q3 Q4 Q5
167 seasonal regression Table 8 . It implies that seasonality in calorie consumption is pronounced, which is consistent with our observation in Fig. 1 and Fig. 2. Most of monthly dummies are statistically significant at 1% level of significance except in Eteya. For households in Eteya, only four of monthly dummies are found to be statistically significant at 10 or 5 percent of level of significance. As most of households in Eteya are relatively better‐off than other villages most of them are in the middle and higher income groups , they can mitigate seasonality in food consumption. For households in D/Brihan and Azedebo, calorie consumption in the month of September is higher than in January. Calorie intake in the months of April is higher than the reference months although statistically insignificant. Similarly, calorie intake in the months of September‐November, and March‐May is higher than in the months of December‐February as it is inflated by calorie consumption in the months of September and April, respectively.
168 Table 7: OLS coefficient estimates of monthly calorie consumption in
kcal/day/person
Full
sample All but
Azedebo D/Birhan Yetmen
Eteya Azedebo
June ‐0.177 ‐0.187 ‐0.528 ‐0.282 0.241 ‐0.149
0.043 ** 0.050 ** 0.066 ** 0.065 ** 0.100 * 0.083
July ‐0.300 ‐0.242 ‐0.383 ‐0.204 ‐0.140 ‐0.476
0.053 ** 0.060 ** 0.088 ** 0.092 * 0.124 0.106 **
August ‐0.429 ‐0.511 ‐0.698 ‐0.667 ‐0.184 ‐0.195
0.052 ** 0.062 ** 0.087 ** 0.098 ** 0.122 0.093 *
September 0.153 0.075 0.158 ‐0.007 0.065 0.376
0.045 ** 0.051 0.064 * 0.078 0.100 0.094 **
October ‐0.250 ‐0.243 ‐0.200 ‐0.453 ‐0.096 ‐0.322
0.050 ** 0.054 ** 0.073 ** 0.077 ** 0.110 0.125 *
November ‐0.277 ‐0.268 ‐0.483 ‐0.227 ‐0.087 ‐0.300
0.048 ** 0.055 ** 0.082 ** 0.078 ** 0.114 0.097 **
December ‐0.445 ‐0.461 ‐0.854 ‐0.635 0.154 ‐0.401
0.055 ** 0.063 ** 0.089 ** 0.102 ** 0.117 0.109 **
February ‐0.508 ‐0.580 ‐0.857 ‐0.589 ‐0.291 ‐0.307
0.053 ** 0.063 ** 0.083 ** 0.088 ** 0.132 * 0.093 **
March ‐0.194 ‐0.152 ‐0.442 ‐0.240 0.215 ‐0.315
0.049 ** 0.055 ** 0.076 ** 0.082 ** 0.108 * 0.101 **
April 0.077 0.081 0.097 0.094 0.062 0.066
0.045 0.054 0.069 0.093 0.109 0.079
May ‐0.350 ‐0.359 ‐0.460 ‐0.422 ‐0.198 ‐0.321
0.046 ** 0.053 ** 0.071 ** 0.087 ** 0.101 0.091 **
Constant 11.803 10.108 7.345 9.235 8.769 15.792
2.006 ** 2.402 ** 5.548 6.516 5.086 3.726 **
F‐ test for months 35.67 27.60 39.08 12.14 6.60 11.68
p value 0.000 0.000 0.000 0.000 0.000 0.000
F statistic 51.3 30.7 28.0 16.2 10.0 13.3
Adjusted R‐squared 0.31 0.26 0.41 0.37 0.21 0.23
Observations 2,806 2,083 727 654 703 724
p 0.1; * p 0.05; ** p 0.01
Note: Robust standard errors in brackets; the regressions were controlled for household total income and its squared, age, sex and education level of household head; household composition and income quintile.
Table 8: OLS coefficient estimates; Log per capita calorie consumption
Full
sample All but
Azedebo D/Birhan Yetmen
Eteya Azedebo
Jun.‐Aug. ‐0.015 0.015 ‐0.006 ‐0.022 0.064 ‐0.102
0.033 0.037 0.051 0.059 0.066 0.071
Sep.‐Nov. 0.168 0.167 0.347 0.154 ‐0.006 0.169
0.036 ** 0.038 ** 0.054 ** 0.052 ** 0.075 0.079 *
Mar.‐May 0.133 0.177 0.205 0.213 0.114 0.004
0.035 ** 0.039 ** 0.064 ** 0.064 ** 0.066 0.072
Constant 12.919 11.752 4.513 15.093 9.577 16.125
2.302 ** 2.759 ** 6.793 7.701 6.861 4.165 **
F test for months 16.90 12.14 33.14 7.30 1.26 6.10
p value 0.000 0.000 0.000 0.001 0.285 0.003
F statistic 44.1 25.0 15.6 14.9 14.1 10.5
Adjusted R‐squared 0.43 0.37 0.44 0.50 0.35 0.28
Observations 960 714 248 220 246 246
p 0.1; * p 0.05; ** p 0.01
Note: Robust standard errors in brackets
169
The regression results are also controlled for per capita income and its squared, per capita livestock ownership in TLU , head sex and education level, household composition age groups as well as expenditure quintiles Table A4 . Most of them, however, found to be statistically insignificant; except for per capita livestock ownership for full sample and for households in D/Brihan and Azedebo. Age composition is significant in most of the cases. As we have indicated in the descriptive section, per capita calorie consumption is significantly greater than for households in the highest quintile 5th quintile .
The Working‐Leser model, equation 2 , is estimated for six food grouping cereals, pulses, root crops, fruits & vegetables, animal products and “other food staffs” Table 9 and 10 . The budget shares in total expenditure is used as dependent variables and regressed against log total food expenditure and it squared, age of household head and its squared, sex of head and family size as well as villages and monthly dummies. The Wald test display existence of seasonality in consumption for cereals, pulses, root crops, animal products and “other foods” in Table 9 but only for cereals and pulses for seasonal dummies Table 10 . The Adjusted R2 values are 0.66, 0.74, 0.84, 0.78 and 0.47, respectively, in the former group Table 9 . Note that food budget share in the month of January reference is lower than other months except in the month of September and April that coincides with Holiday festivals. Table A5 and A6 present the full regression results. Log per capita consumption has positive and significant impact for each food group. Log total cultivated land has positive impact on the share of cereal consumption but negatively influence other food groups. More lands are allocated for cereal production as its share mount in the total household budget. Age and sex of head